Research On DOA Estimation Based Downlink Beamforming Techniques For FDD System

Smart antenna technology is one of the key technologies of the thirdgeneration mobile communication system. And it introduces space divisionmultiple access which makes use of different transmission paths ofdifferent signals to differentiate the signals from the same frequency, thesame time slot and the same code. Moreover, space division multiple accesscan also be combined with other multiple access technologies, such as timedivision multiple access, code division multiple access, and so on. Smartantenna technology makes the range of communication resource extendfrom time region, frequency region and code region to space region. Thussmart antenna can remarkably increase the capacity of communicationsystem, and sufficiently exploit limited frequencY spectrum resource. In anFDD wireless system, the duplex spacing between uplink and downlink islarger than the coherence bandwidth of the mobile radio channel, thus theuplink and the downlink small-scale fading is uncorrelated. Moreover, thearray response is correlated with the frequency of the signal carrier. So it isimpossible to resue the uplink weights for downlink beamforming. As aresult, the application of smart antenna for the downlink is a difficult problem. However, if the duplex spacing between the uplink and thedownlink isn't large enough, the uplink channel parameters like thedirections-of-arrival can be used for the downlink beamforming. This paperplaces its importance on the research of DOA estimation algorithms.The major works of this paper can be summarized as follows:1. This paper establishes the uplink and downlink channel model.According to this model, it knows how to calculate DCCM.2. Many high resolution subspace-based methods like MUSIC andESPRIT estimate the direction of arrival of plane waves impinging on anantenna array via eigendecomposition in array signal processing. However,high computational burden for eigendecomposition makes them unsuitablefor real time processing. Moreover, MUSIC algrithom can't estimate weaksignals very well, and it is verified through the simulation.3. This paper presents a DOA estimation method based on a novelprincipal component analysis technique. It introduces an iterative approachfor estimating the signal subspace to MUSIC algorithm, which substitutesthe steps that obtain signal subspace via eigendecomposition in originalalgorithm. Simulation results show that the proposed algorithm has betterperformance than MUSIC algorithm when estimating weak signals and itscomputational complexity is smaller, which can be seen from the learningcurves.4. On the basis of DOA estimation, this paper integrates the mainideas in literatures[47-50] firstly, and then deduces them in detail, after thatthis paper achieves the initial weight according to those results deduced.Based on these works finished before, aiming to minimize the arraytransmission power, it optimizes the initial weight and gains optimized weight. In the end, through the simulation this paper compares their effectson the beam pattern when initial weight and optimized weight are used forthe downlink beamforming.